Accounting for non-normal distribution of input variables and their correlations in robust optimization

Abstract In this work, metamodel-based robust optimization is performed using measured scatter of noise variables. Principal component analysis used to describe the input linearly uncorrelated principal components. Some these components follow a normal probability distribution, others however deviate from distribution. that case, for more accurate description material scatter, multimodal distribution used. An analytical method implemented propagate via metamodel and calculate statistics response accurately efficiently. The criterion as well constraints evaluation are adjusted properly deal with response. Two problems presented show effectiveness proposed approach validate method. A basketball free throw in windy weather condition forming B-pillar presented. significance accounting non-normal variables distributions investigated. Moreover, calculation statistics, adjustment problem discussed.